roi.barkan@gmail.com More Ranges Please CppCon 2023 1 Slidesroi.barkan@gmail.com Hi, I’m Roi ● Roi Barkan (he/him) - ןקרב יעור ● I live in Tel Aviv ● C++ developer since 2000 ● SVP Development questions, make comments 2 Slidesroi.barkan@gmail.com Outline ● Libraries ○ What ○ Why ○ How ● Ranges ○ Introduction ○ Strengths, core ideas ● Rabbits ● Summary 3roi.barkan@gmail.com Libraries on Science link STL and its Design Principles 2002 11roi.barkan@gmail.com C++ Ranges 12roi.barkan@gmail.com Ranges is a Breakthrough Library 13 ● One of C++20 big-four features ● Rests on decades

E�ective Ranges: A E�ective Ranges: A tutorial for using tutorial for using C++2x ranges C++2x ranges Je� Garland Je� Garland Created: 2023-10-03 Tue 08:50 1Intro Intro the beginning of the end – – for begin and end 2 . 1hello ranges hello ranges 2 . 2the old way: the old way: sort sort std::array a { 6, 2, 3, 4, 5, 1 }; std::sort ( a.begin(), a.end() ); for ( auto v : a ) cout cout << v << " " << endl; 2 . 3the ranges way: the ranges way: sort sort namespace rng = std::ranges; std::array a { 6, 2, 3, 4, 5, 1 }; rng::sort ( a ); //clear, obvious meaning, -13 characters

Repository Maintainer 23TALK OVERVIEW Brief introduction to Particle Filters Brief mention to the C++ Ranges library Implementation walkthrough of a Particle Filter using C++23 Practical recommendations and etc.) 0:00 / 0:28 7C++ RANGES LIBRARY Extension and generalization of the algorithm and iterator libraries. (C++98) algorithm(begin(range), end(range)) ➡ result 8C++ RANGES LIBRARY Extension and generalization adaptor(range) ➡ view (C++20) range | adaptor ➡ view (C++20) range | adaptor1 | adaptor2 ➡ view 8.1C++ RANGES LIBRARY Extension and generalization of the algorithm and iterator libraries. (C++98) algorithm(begin(range)

return reduce(par_unseq, z, 0, std::plus()); }Outline - Background (Ranges, Parallelism, Distributed Data Structures) - Distributed Ranges (Concepts) - Implementation (Algorithms and views) - Complex (Dense and sparse matrices) - Lessons learnedOutline - Background (Ranges, Parallelism, Distributed Data Structures) - Distributed Ranges (Concepts) - Implementation (Algorithms and views) - Complex views of data - Algorithms - Operate on and modify data using namespace std; using namespace std::ranges; using namespace std::execution; template auto dot_product(R&& x, R&& y) { using T =

std::same_as; }; template concept GravitationalRange = GravitationalBodyranges::range_value_t>; 1 2 3 4 5 6 7 8 9 10 78910def GetForceOnBody(this_body, all_bodies): singleForceOnThisBody); 15 } 16 12131415What is a view? 17Easy to access: namespace std { namespace views = ranges::views; } 1 2 3 18Range View Semantically: a collection of "things". Has begin and end functions construction. ranges::enable_view is true. template concept range = requires(T& t) { ranges::begin(t); ranges::end(t); }; template concept view = ranges::range &&

2024-09-19 13New std::ranges algorithm Old: std::transform(std::begin(rng), std::end(rng), std::back_inserter(squares), [](int x) { return x * x; }); New: namespace stdr = std::ranges; stdr::transform(rng Functional Programming in C++ CppCon 2024-09-19 29Composition shape: Multiple ranges → single range Input Multiple ranges Output Single range Jonathan Müller — @foonathan Back to Basics: Functional Functional Programming in C++ CppCon 2024-09-19 30Composition shape: Multiple ranges → single range Input Multiple ranges Output Single range stdv::cartesian_product: std::tuple of all possible element combinations

Value-based TMP - find_index - template-less refactor constexpr auto find_index(auto t, const std::ranges::range auto& ts) -> std::size_t { } 18 / 58Value-based TMP - find_index - template-less refactor Value-based TMP - find_index - template-less refactor constexpr auto find_index(auto t, const std::ranges::range auto& ts) -> std::size_t { for (auto i = 0u; i < ts.size(); ++i) { if (ts[i] == t) Value-based TMP - find_index - template-less refactor constexpr auto find_index(auto t, const std::ranges::range auto& ts) -> std::size_t { for (auto i = 0u; i < ts.size(); ++i) { if (ts[i] == t)

collected3ChatGPT 4ChatGPT 5Median #include #include <ranges> #include auto median(std::vector& v) { std::ranges::sort(v); const auto medianIndex = std::size(v) / 2; return 7Goal 8 template ranges::random_access_range Range> constexpr auto median(Range&& rg) { // TODO // O(N)? }Stretch Goal 9 template ranges::random_access_range Range Range, typename Comparison = std::ranges::less, typename Projection = std::identity> constexpr auto median( ExecPolicy&& execPolicy, Range&& rg, Comparison comp = {}, Projection proj = {}) {

*/ ExampleBringing parallelism to std::ranges algorithmsISO C++ Parallelism/Concurrency Programming Language (based on Gonzalo’s ISC C++ BoF) C++ 2023 Ranges++ many Multi-dimensional Spans [Bryce, operator[i, j, k] C++ 2026 Executors/Sender/Receiver [P2300] Reflection Algorithms++ asynchrony, ranges [Ruslan, Alexey, Bryce] Linear Algebra, submdspan, padded layouts [Bryce, Christian] Concurrency++ Forward Progress C++ 2020 Concepts Ranges Modules Concurrency++ [Bryce, Gonzalo] Coroutines atomic_ref, barriers, … 2024Bringing parallelism to std::ranges algorithms (P3179) // C++03 template

0 码力 | 56 页 | 514.85 KB | 5 月前

3

algorithms were a thing When we realised they were important after all STL algorithms Modern C++ ranges4 auto const inputs = std::vector{1, 2, 3, 4, 5, 6, 7, 8, 9, 10}; auto results = std::vector{}; Filtering transform_if ?5 RANGES transform_if auto const inputs = std::vector{1, 2, 3, 4, 5, 6, 7, 8, 9, 10}; std::vector const results = inputs | ranges::view::filter([](int i){ return 0; }) | ranges::view::transform([](int i){ return i * 2; }); auto const inputs = std::vector{1, 2, 3, 4, 5, 6, 7, 8, 9, 10}; std::vector const results = inputs | ranges::view::transform([](int